Sie sind auf Seite 1von 13

High temperature solar thermal central-receiver billboard design

Nicholas Boerema
a,
, Graham Morrison
a
, Robert Taylor
a
, Gary Rosengarten
a,b
a
School of Mechanical and Manufacturing Engineering, University of New South Wales, Sydney, NSW 2052, Australia
b
School of Aerospace, Mechanical and Manufacturing Engineering, RMIT University, Melbourne, Vic 3001, Australia
Received 28 January 2013; received in revised form 19 August 2013; accepted 3 September 2013
Available online 24 September 2013
Communicated by: Associate Editor Ranga Pitchumani
Abstract
The design of central receivers in solar thermal power plants is critical for ecient plant operation and sucient operational lifetimes.
The high, non-uniform concentration ratios used in central receivers lead to high, non-uniform receiver temperatures. For the same oper-
ational conditions, small changes to the receiver design can make a big impact on the expected lifetime of the receiver. This is due to
limitations of the receiver materials to high temperatures and thermal cycling. In this study, we investigate the eect of several engineer-
ing concepts on the resultant surface temperatures of tubular billboard receivers. Four tubular billboard designs are investigated along
with the sensitivity these designs have to high temperatures resulting from changes in the aiming point of the heliostat array. We exam-
ined a receiver with single diameter tubes, an ideal ow receiver, a receiver using various diameter tubes and a receiver made of tube
panels in series. The single-diameter and multi-diameter receivers were found to have high temperatures and high sensitivity under
non-standard irradiation. The multi-pass receiver was found to out-perform the other designs by reducing both the maximum surface
temperatures under standard irradiance and the risk of high temperatures from irradiance changes. The results provide insights into
tubular billboard receiver design, material selection and design for extended life.
2013 Elsevier Ltd. All rights reserved.
Keywords: Concentrated solar thermal; Central receiver system; Receiver design; Billboard
1. Introduction
Solar thermal Central Receiver Systems (CRSs) are
expected to provide a path for achieving large-scale deploy-
ment of electricity generators that use a renewable resource
(IEA, 2010). CRSs consist of a eld of heliostats which
focus solar radiation towards a receiver, situated at the
top of a tower. Solar energy is collected in the receiver
using a heat transfer uid (HTF) which is then used to
transfer the energy to a thermodynamic cycle (via steam)
to allow electricity generation. CRSs may consist of a large,
single tower design (such as used at Gemasolar (Dunn
et al., 2012)), or multiple towers may be used, such as in
the eSolar design (Schell, 2011).
Focusing the reected light from the heliostats towards
a point means that high solar concentration ratios can be
achieved on the receiver. High concentration ratios mean
high receiver eciencies, and allow higher working cycle
temperatures to be achieved which leads to higher thermo-
dynamic eciencies. Increasing the eciency is important
for CRSs, as it reduces the required size of the heliostat
eld for the same generation capacity. This is crucial to
economic viability since the heliostat eld makes up
approximately 40% of the systems capital costs (Hinkley
et al., 2011). To further increase eciency, small receiver
areas are desired as this reduces receiver losses from con-
vection and re-radiation (Bignon, 1980). A challenge for
CRSs is to increase the durability of the highly irradiated
receiver, whilst allowing high temperatures to be achieved
and ensuring a cost eective design.
Whilst cavity and volumetric receivers are receiving a lot
of attention from the academic community (Romero et al.,
0038-092X/$ - see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.solener.2013.09.008

Corresponding author. Tel.: +61 428828015.


E-mail address: nicholas.boerema@student.unsw.edu.au (N. Boere-
ma).
www.elsevier.com/locate/solener
Available online at www.sciencedirect.com
ScienceDirect
Solar Energy 97 (2013) 356368
2002; A

vila-Mar n, 2011), it is external tubular receivers


that are being used most in commercial central receiver
projects (NREL, 2011; IT Power, 2012). This results from
tubular receiver technology being built on concepts from
heat exchanger design i.e. relatively inexpensive, durable,
proven technology.
Currently, the majority of commercial systems being
deployed are using a surround tower heliostat eld which
focus incident light onto a cylindrical tubular receiver.
Modelling of heliostat layouts has shown that equator fac-
ing heliostat elds can greatly increase the eld optical e-
ciency, which reduces the size of the heliostat eld required
compared to a surround tower design (Collado, 2009;
Schell, 2011). Equator facing elds have a lower mirror area
to number of towers ratio, which means that they are more
suited to a plant design that uses multiple smaller towers.
A result of multiple smaller towers is that the cost of the
receiver becomes more critical due to the number required.
The importance of simple operation is also increased for the
same reasons. For an equator facing heliostat eld a cylin-
drical receiver is no longer a suitable design. Instead, a suit-
able option is a billboard receiver, which consists of a group
of vertical tubes, aligned in a single plane (Eduardo and
Manuel, 2007). This receiver design and eld layout, how-
ever, can result in highly non-uniform ux densities, which
can introduce high surface temperatures. As the thermal
stresses are related to the surface temperatures, it is neces-
sary to predict and control these temperatures.
Tracking error is inherent in all heliostats, due to sun
positioning error, referencing errors, gravity sag, pedestal
tilt, mirror and support vibration (wind impacts), cant-
ing/bore sight errors and gear backlash. These errors can
be reduced, but in general this incurs an increase in helio-
stat and system costs (Zhang et al., 2012). This error leads
to a probability distribution of where the reected light will
land on the receiver for a given aiming point. Due to this
distribution it is undesirable to focus the heliostats towards
the edges of a billboard receiver as this increases spillage
(where incident irradiance misses the receiver). This can
cause issues for billboard receivers as the incident irradi-
ance is highly concentrated towards the centre of the recei-
ver. This non-uniform ux distribution can lead to very
high surface temperatures. As the thermal stresses are
related to the surface temperatures, it is necessary to pre-
dict and control the receiver surface temperatures. If tem-
peratures are known, receiver materials and design can be
selected appropriately to handle the stresses over its
20,000100,000 h design operational life.
In selecting a material for the receiver tubes, this high
number of thermal cycles must be considered. The number
of thermal cycles needed to be withstood over the tube life-
time is environment specic but will be in the tens to hun-
dreds of thousands. This thermal cycling means that the
issues encountered in conventional heat exchanger designs
will be exacerbated for the same maximum temperature
limits of operation. Furthermore the non-isothermal condi-
tions across the receiver mean that multiple temperature
regions must be considered. Signicant issues to consider
are loss in ductility and tensile strength, creep and carburi-
zation rates, fatigue crack growth rates, and oxidation
resistance. Fig. 1a demonstrates a loss in material strength
resulting from high service temperatures whilst Fig. 1b
shows the clear trend in increasing crack growth rate with
increasing temperature. Together, the gures demonstrate
the need to understand the expected temperatures that
are likely to result for a given receiver design.
To assist in this understanding, an assessment of the
resultant surface temperatures has been performed for four
dierent billboard designs; a multi-pass receiver, a
single-diameter receiver, an ideal ow receiver, and a
multi-diameter receiver.
1.1. Receiver concepts
The single-diameter billboard design (Fig. 2) is the sim-
plest of the designs as it uses only a single panel made of
Nomenclature
Acronyms
CRS central receiver system
HTF heat transfer uid
DNI direct normal irradiance
Symbols
Nu Nusselt number
Re Reynolds number
Pr Prandtl number
T
1
HTF temperature at start of step
T
2
HTF temperature at end of step
_
Q
f
heat transfer to segment of HTF (W)
_ m mass ow rate (kg/s)
C
p
HTF specic heat at constant pressure (J/kg K)
h
combined
enthalpy of the combined HTF from all recei-
ver pipes in parallel (J/kg)
_ m
combined
combined HTF from all receiver pipes in par-
allel (kg/s)
h
k
enthalpy of HTF at exit of k
th
receiver pipe (J/
kg)
_ m
k
mass ow rate of k
th
receiver pipe (kg/s)
DP pressure drop (Pa)
f friction factor
L receiver pipe length (m)
D receiver pipe internal diameter (m)
q density (kg/m
3
)
V velocity (m/s)
N. Boerema et al. / Solar Energy 97 (2013) 356368 357
uniform diameter, parallel pipes. This design operates with
a uniform ow rate through each pipe. Thus, for a non-uni-
form irradiance prole, the exit temperature of the HTF in
the central pipes is much higher than the outer tubes to
allow a suciently high mean temperature after all the uid
mixes. As the surface temperature of the receiver will be
above the HTF temperature, high surface temperatures will
result. This places the receiver under thermal strains which
standard stainless alloys may not be able to handle for the
length of the receivers design life (Ward, 2012). To mini-
mise receiver costs it is desirable to be able to use low-cost
stainless steels, and thus methods for reducing the maxi-
mum temperature of this receiver design need to be
developed.
One way to reduce surface temperatures is to ensure that
the HTF from all pipes reaches the desired outlet tempera-
ture. This could be achieved by either using ow control
devices or through varying the pipe geometry to match
the heat ux on each pipe. An ideal ow receiver is
proposed here (Fig. 3a), in which the ow rate of each
individual tube is adjusted such that the uid outlet tem-
perature for each tube is equal to the desired exit tempera-
ture for any ux distribution.
A multi-diameter receiver is proposed (Fig. 3b) to par-
tially achieve this. This receiver uses a selection of multiple
diameter pipes-each with a single diameter to make up
the receiver. As the pipes are in parallel the pressure drop
along each pipe will be equal. The mass ow rate is then
proportional to the pipe ow resistance only. As the pres-
sure drop is constant between pipes, the smaller diameter
pipes have a higher resistance to ow and thus a lower
mass ow rate which will heat up more easily. Thus, for
a Gaussian ux distribution; by using smaller diameter
pipes towards the edge of the receiver, where irradiance is
lower, outlet temperatures closer to that desired can be
achieved.
The multi-pass tubular billboard design (Fig. 3c) is
made up of multiple side-by-side panels through which
the HTF passes in both series and parallel, beginning in
the outer panels and nishing by passing through the centre
most panel (Schiel and Geyer, 1988). The HTF may be
combined in a header between panels. As the irradiance
is greatest towards the centre, this design allows the HTF
temperature to be better matched to the irradiance level.
The main issue with this receiver design is that it is not pos-
sible to completely drain the receiver of the HTF without
using additional valves. This leads to undesirable complex-
ity and additional costs. Whilst it is beyond the scope of
this paper, multiple diameters could also be incorporated
into this design.
2. Model development
To determine the surface temperatures that will result
from a given receiver ux distribution, a heat transfer
model of a tubular billboard receiver has been developed.
For the model, the ux distribution, the inlet temperature
and desired outlet temperature are used as inputs. As out-
puts, the necessary ow rate and resultant heat transfer and
surface temperatures are solved. The model allows for
comparison between single-diameter, multi-diameter and
multi-pass receivers.
(a) (b)
Fig. 1. (a) Creep rate curves for several austenitic stainless steels: 1% creep in 100,000 h (Davis, 2001). (b) Variation of fatigue crack growth rate as a
function of temperature at stress intensity factor = 30 MPa

m
p
(Viswanathan, 1989). Da/dN indicates the increase in crack length for each cycle N, of
sinusoidally varying applied stress.
Fig. 2. Schematic of conventional single pass tubular billboard receiver.
358 N. Boerema et al. / Solar Energy 97 (2013) 356368
For the receiver model, sodium has been taken as the
heat transfer uid, using equations for the thermophysical
properties and pressure drop as described by Fink and
Leibowitz (1995) and Boerema et al. (2012). The following
equation for calculating the Nusselt number in turbulent
ow has been used and is recommended by Cengel (2007)
for liquid metals due to their very low Prandtl numbers.
Nu 6:3 0:0167Re
0:85
Pr
0:93
s
1
where Re is the Reynolds number and Pr
s
is the Prandtl
number calculated using properties at the pipe inner sur-
face temperature. For Eq. (1) the ow is assumed to be tur-
bulent and fully developed. The assumption of turbulence
was seen as valid due to Reynolds numbers calculated
being on the order of 10
4
. The assumption of fully devel-
oped ow was seen as valid as normally a tubular receiver
will have some unexposed tube before the irradiated sec-
tion, which would provide sucient length for the ow to
become fully developed.
Sodium has been used as the heat transfer coecient due
to its high thermal conductivity which makes the resultant
heat transfer coecients less dependent on the ow condi-
tions. Other heat transfer uids would have similar surface
temperature distributions for a given receiver concept,
however the tube lengths would need to be adjusted to
ensure that adequate ow rates were achieved to insure suf-
cient heat transfer. Surface fouling has been assumed to
be negligible.
2.1. Receiver ux distribution
The ux distribution on the target depends on the par-
ticular heliostat design, the eld layout and the time of
day. In this study a heliostat eld with a single point aiming
strategy has been used. The resulting ux distribution
across the receiver aperture has been approximated by a
normal (Gaussian) distribution, with a mean located at
the centre of the receiver. The standard deviation of the dis-
tribution will depend namely on the referencing errors,
gravity sag, pedestal tilt and canting/bore sight errors,
the amount of backlash in the gears, mirror errors and
environmental conditions (wind). The ux will be distrib-
uted about the exposed section of the perimeter for each
of the receiver pipes. For the exposed section, taken as
180 of pipe for this study, the distribution about the
perimeter is approximately given by q(h) = q
net
cos (h),
where q
net
is the intensity of the incident radiation (see
Fig. 4) (Yang et al., 2012).
A 1.5 m 1.5 m receiver made up of parallel tubes has
been used. For the ux distribution, a mean of zero and
Fig. 3. (a) Ideal ow receiver. (b) Multi-diameter receiver. (c) Multi-Pass receiver.
Fig. 4. Flux distribution (red) about the tube perimeter. (For interpreta-
tion of the references to colour in this gure legend, the reader is referred
to the web version of this article.)
N. Boerema et al. / Solar Energy 97 (2013) 356368 359
a standard deviation of 0.255Width has been chosen,
resulting in approximately 10% spillage and leading to an
average incident irradiance of 737 kW/m
2
across the recei-
ver aperture. To create the distribution (Fig. 5), the receiver
aperture was rst divided into a mesh, with an integer num-
ber of segments per receiver tube. The spillage fraction in
the vertical and horizontal directions was then decided
and used respectively to calculate the standard deviation
for the ux distributions for the two directions. A lower
spillage fraction results in a lower standard deviation for
the ux distribution and thus a higher peak ux. Matlabs
multivariate normal cumulative distribution function was
then used to calculate the multivariate normal cumulative
probability evaluated over each segment. This gives the
expected percentage of incident energy landing on each seg-
ment. This fraction was then multiplied with the total
energy coming from the eld to give the energy landing
on each segment.
2.2. Heat transfer calculations
One-dimensional heat transfer calculations have been
performed on the receiver by dividing each pipe into 10
evenly spaced segments across its diameter and into seg-
ments of length 50 mm along the pipes main axis. To start
the calculations, an initial mass ow rate is assumed. Using
the uid inlet temperature, the uid properties and the fric-
tion factor, the heat transfer coecient can be calculated
for the rst segment. The pipe wall thermal resistance,
the thermal convective resistance and the total resistance
can be calculated using standard one dimensional pipe
thermal resistance equations (Cengel, 2007, pp.146152).
The segments surface temperature can then be solved
by simultaneously solving standard heat transfer equations
for conduction/convection to the HTF and radiative and
convective heat losses to the surroundings, such that the
combined energy transferred is equal to the absorbed
energy. This also solves the energy transferred to the
HTF for each segment and the thermal losses of each
segment. The energy transferred to the HTF for each seg-
ment of a single step along the pipes main axes can then
be summed together to give the total energy transferred
to the HTF for that step,
_
Q
f
.
The convective heat transfer coecient is dependent on
environmental conditions and as such a reference value of
16 W/m
2
K has been used. This relates to a wind speed of
approximately 4 m/s (Kesselring and Selvage, 1986, pp.
5.724). For simplicity, the view factor between the pipes
and the surrounds has been approximated as cos h. A
reectivity of 0.08 has been used for the Pyromark surface
coating. This value has been adjusted by a correction factor
of 0.72 to account for the reectivity of light o of the pipes
onto other pipes, giving a corrected reectivity of 0.0576
(thus absorptivity = 0.9424). The correction factor was
approximated from calculations made for this geometry
by Kesselring and Selvage (1986, pp. 5.19).
Assuming sucient mixing, the sum of the heat transfer
to the HTF,
_
Q
f
, for all segments in a single step along the
pipes main axis, can then be used to calculate the temper-
ature at the end of the step (T
2
) and is given by:
T
2

_
Q
f
_ mC
p
T
1
2
T
1
is the temperature at the start of the step and is equal to
the inlet temperature for the rst step. _ m is the mass ow
rate through a single pipe and C
p
is the specic heat of
the uid, calculated as a function of temperature. These
calculations can be repeated for each pipe in the receiver.
The exit temperature of the combined uid leaving the re-
ceiver pipes (i.e. the header temperature) can be solved for
by using the enthalpy of the combined HTF:
h
combined

1
_ m
combined
X
n
k1
_ m
k
h
k
J=kg 3
where _ m
combined
is the combined mass ow rate and n is the
total number of pipes. _ m
k
and h
k
are the mass ow rate and
enthalpy, respectively, for the k
th
pipe. The process can be
Fig. 5. Flux distribution across the receiver aperture. Mean = 0, standard deviation = 0.255

Width.
360 N. Boerema et al. / Solar Energy 97 (2013) 356368
iterated, adjusting the mass ow rate until the exit temper-
ature of the combined uid leaving the receiver pipes is
equal to the desired exit temperature. To match with com-
mon tower operational conditions, a desired exit tempera-
ture of 570 C has been used (Kolb, 2011).
2.3. Validation and verication
To verify the model for the dierent receiver designs var-
ious predictable tests were run and the results compared
with those expected. These included a uniform ux distri-
bution, and zero loss cases-where all incident energy is
absorbed and the thermal emissivity and coecient of heat
transfer were reduced to zero. The uniform ux distribu-
tion meant that the surface temperatures were identical
for each tube and increasing approximately linearly with
uid temperature.
Grid convergence studies were performed on the model
to ensure that an adequate mesh density has been chosen.
Various predictable cases, such as uniform intensities and
no losses were also performed to verify the model. For sim-
plicity, conduction along and around the tubes was
neglected, as was absorption of reradiated light from neigh-
bouring tubes. These heat transfer mechanisms would act
towards levelling temperatures across the receiver, how-
ever, the energy transferred is minimal relative to the
incoming ux, due to the low pipe-to-pipe view factor
and low thermal conductivity and cross-sectional area of
the tube.
The results of the model were validated through com-
parisons with results from the IEASSPS sodium receiver
and the thermodynamic model HOTREC developed
and validated as part of the IEASSPS program (Schiel
and Geyer, 1988). The SSPS sodium receiver was a bill-
board receiver which consisted of 5 panels each made up
of 39 tubes of 14 mm diameter and 1 mm wall thickness.
The sodium was passed through each of these panels in ser-
ies. The order through which the panels were passed can be
seen in Fig. 6. Results from testing of the receiver and from
the thermodynamic model were presented by Schiel and
Geyer (1988) for both a triple point aiming strategy and
a single point aiming strategy. In the triple point aiming
strategy a third of the light from the heliostat eld was
focused onto each of the three central panels. In the single
point aiming strategy all incident light was focused onto
the central panel. HOTREC results presented were devel-
oped using 2.76 MW and 2.8 MW for the incident power
in the triple and single aim points respectively. The peak
ux for the triple aiming point strategy was 1.6 MW/m
2
,
whilst for the single aiming point strategy the peak ux
was 2.3 MW/m
2
. To allow validation of our model it was
attempted to match the ux and the peak uxes used in
the IEASSPS project such that the HTF, pipe outer sur-
face and Pyromark surface coating temperatures could be
compared with those presented. In the vertical direction
the ux is presented for the central pipe. For the horizontal
direction the absorbed energy in each panel was presented
whilst panel eciencies were presented by Kesselring and
Selvage (1986). The absorbed energy, however, also
included the eect of 30 broad focused heliostats, which
added another 0.6 MW to the total power and 0.1 MW/
m
2
to the peak ux. These values were used to calculate
the fraction of incident energy on each panel for the HOT-
REC model, which was then attempted to be achieved with
our ux, whilst still maintaining the correct overall incident
energy and peak ux. It should be noted that an exact
match of the ux was not achievable due to the number
of variables and the limited information presented in the
Schiel and Geyer paper and as such there is some misalign-
ment with the temperatures and incident energy on each
panel.
Figs. 7 and 8 allow a comparison between the HOTREC
model and our model for the temperatures of the central
tube of the fth panel for both the single aiming point
strategy (Fig. 7) and the triple aiming point strategy
(Fig. 8). Considering the ux mismatch, the temperature
results show very good agreement. This agreement with
the grid convergence tests gave us condence in the validity
of our model.
Table 2 shows the percentage of incident energy for each
panel, used to help create the ux prole, and the resultant
eciencies for each panel. Comparing the eciencies for
the triple point aiming strategy they are also in good agree-
ment, considering the ux mismatch and number of
parameters.
Fig. 6. Schematic representation of the ow path for the advanced sodium
receiver at IEASSPS.
N. Boerema et al. / Solar Energy 97 (2013) 356368 361
3. Receiver surface temperatures
To compare the dierent designs a consistent set of
receiver variables have been used, as outlined in Table 1.
3.1. Centred ux distribution
3.1.1. Single-diameter receiver
The resultant surface irradiance and surface tempera-
tures of a single-diameter billboard receiver has been deter-
mined. Tubes with an outer diameter of 25.4 mm, a length
of 1.5 m and a wall thickness of 1 mm have been used. The
ux distribution across the receiver tubes is presented in
Fig. 9. The thermal eciency for this receiver under the
modelled conditions is 91.2%, where receiver eciency is
dened as energy absorbed by the HTF to energy incident
on the receiver (and thus does not include spillage).
Fig. 10 shows the resultant surface temperatures indicat-
ing a maximum pipe surface temperature of 861 C when
trying to achieve an outlet temperature of 570 C. The
low irradiance levels on the pipe surfaces towards the edge
of the receiver result in a low exit temperature for those
pipes. To compensate, the inner pipes must have a uid exit
temperature well above the desired operating temperature
Fig. 7. Temperature and heat ux proles along the central tube of the fth panel as calculated by HOTREC and our model for the single point aiming
strategy with incident power = 2.8 MW and peak ux = 2.3 MW/m
2
.
Fig. 8. Temperature and heat ux proles along the central tube of the fth panel as calculated by HOTREC and our model for the triple point aiming
strategy with incident power = 2.76 MW and peak ux = 1.6 MW/m
2
.
362 N. Boerema et al. / Solar Energy 97 (2013) 356368
(Fig. 11). As the surface temperature will be above the uid
temperature, the result of this is high surface temperatures
towards the centre of the upper portion of the receiver.
An issue in this receiver design is that if the aiming point
for all heliostats moves away from the centre of the recei-
ver, very high surface temperatures will result if the ow
rate is adjusted such that the desired operating temperature
is still achieved. To reduce this risk, control which ensures
minimum ow rates could be used. These ow rates could
be based on the suns direct normal irradiance (DNI) level
and the number of heliostats focused on the receiver. For
greater protection a camera would likely be needed,
focused on the receiver. The images from the camera would
then allow imaging software to determine the mean of the
irradiance distribution and the heliostat aiming could be
adjusted accordingly as is done by BrightSource Energy
(Bobinecz, 2012). Whilst these methods exist to limit high
temperature occurrences, the particular temperature stabil-
ity of each receiver design under rapid irradiance uctua-
tions, and shifts in the focus point should be understood
so that the risks can be designed for accordingly.
Tubular receivers can also be used in a cavity, as a cavity
receiver. The aim of this is to reduce losses from convection
and re-radiation and to try and achieve an even ux inten-
sity across the receiver surface. This is achieved through re-
radiation inside the cavity and through the incident light
being reected o of the cavity walls. For comparison with
the other receiver designs the uniform diameter receiver has
also been modelled with an even ux distribution of
737 kW/m
2
. The maximum surface temperature seen is
632 C and the resultant receiver eciency is 91.8%. Due
to the even distribution of the light, the surface tempera-
tures are the same for all pipes which will result in lower
thermal stresses.
Table 1
Parameters used in the model.
Example parameters Value
Pipe thickness 1 mm
Pipe thermal conductivity 20 W/mK
Pipe thermal expansion coecient 17.3 10
6
Pipe roughness 0.002 mm
Receiver pipe length 1.5 m
Receiver width 1.5 m
Reectivity 0.08
Corrected reectivity 0.0576
Emissivity 0.85
Surface heat transfer coef. 16 W/m
2
K
Inlet temperature 200 C
Desired outlet temperature 570 C
Flux standard deviation 0.389 m
Spillage 10%
Ambient temperature 20 C
Total ux 1.68 MW
Ave. irradiance on Receiver 737 kW/m
2
Table 2
Incident energy and eciencies for each panel for validation.
Incident Energy Eciency
Single point aiming Triple point aiming Single point aiming Triple point aiming
Hotrec (%) Ours (%) Hotrec (%) Ours (%) Hotrec Ours (%) Hotrec (%) Ours (%)
Panel 1 2 3 4 4 N/A 78 84 81
Panel 4 21 24 27 27 N/A 90 91 90
Panel 5 49 45 33 36 N/A 90 90 89
Panel 3 26 24 29 28 N/A 91 92 91
Panel 2 2 3 6 5 N/A 78 84 81
Fig. 9. Flux distribution across the receiver tube surfaces.
Fig. 10. Pipe surface temperatures to achieve a desired outlet temperature
of 570 C.
N. Boerema et al. / Solar Energy 97 (2013) 356368 363
3.1.2. Ideal ow receiver
The energy balance for the ideal ow receiver can be
solved as before, however, this time the mass ow rate is
adjusted for each individual pipe until the exit temperature
is equal to the desired outlet temperature.
In Fig. 12 it can be seen that the maximum surface tem-
perature has now been reduced to approximately 634 C, a
reduction of over 200 C. The eciency achieved was
91.6%. As each pipe now has a similar temperature distri-
bution the thermal expansion of each pipe will also be sim-
ilar. This will reduce the thermal strains on the receiver.
Fig. 13 shows the mass ow rate of each pipe normalised
with the ow rate of the centre pipe. For equal uid exit
temperatures the ow rates need to be reduced to this frac-
tion. Methods to actively control individual ow rates may
be expensive, however, there may be cheaper passive
temperature actuated options.
3.1.3. Multi-diameter receiver
Flow control can be locked in using dierent pipe diam-
eters at dierent parts of the receiver to control the ow
resistance. Calculations for the multi-diameter receiver
must solve for the mass ow rate of each pipe such that
the pressure drop through all pipes is equal and the desired
header temperature is achieved.
Using properties at the average uid temperature, the
pressure drop can be calculated using the DarcyWeisbach
equation, with the friction factor calculated using the expli-
cit equation developed by Haaland (1983).
Taking one pipe as the reference pipe, the mass ow
rates of the other pipes can be adjusted such that the pres-
sure drop for each pipe is the same as the reference pipe.
The ow rate of the reference must then be adjusted to
ensure that the exit temperature of the combined uid
remains equal to the desired exit temperature.
Fig. 14 shows receiver surface temperatures for a receiver
using 11 dierent pipe diameters (evenly spaced from12 mm
to 25.4 mm). As can be seen, the maximum surface temper-
atures have been reduced (maximum temperature: 707 C).
The reduction in temperatures is, however, far from that
achieved with uniform outlet temperatures (634 C). The
eciency for this receiver under the stated conditions is
92.6%. This eciency is higher than the even ux receiver
as the side tubes allow collection of the incident energy,
whilst not needing to achieve the desired outlet temperatures
(and thus they maintain a low surface temperature). The
lower than desired outlet temperatures can be easily bal-
anced by the higher volume and more highly irradiated pipes
near the horizontal centre of the receiver. In eect, the recei-
ver surface area with high temperatures is reduced without
increasing the maximum temperature signicantly.
3.1.4. Multi-pass receiver
In order to reduce surface temperatures passively we
have considered, a multi-pass receiver. The receiver is made
Fig. 11. Dierence between the HTF exit temperature and the desired exit temperature for a standard billboard receiver with a centred normal heat ux
distribution with a mean irradiance of 737 kW/m
2
.
Fig. 12. Pipe surface temperatures (C) for a receiver using individually
adjusted mass ow rates for each pipe.
364 N. Boerema et al. / Solar Energy 97 (2013) 356368
of pipes all with the same diameter, with the uid rst pass-
ing through a quarter of the tubes on each side of the recei-
ver. The uid is then combined in a header before passing
through the remaining tubes. The results for the surface
temperatures can be seen in Fig. 15. The results show a
maximum surface temperature of 695 C with the rst
header temperature of 307 C.
The eciency for this receiver under the modelled con-
ditions is 92.0%, a 0.8% eciency increase compared to
the single diameter receiver. As with the multi-diameter
receiver, this small increase in eciency is due to the
decrease in area with high temperatures for a relatively
small increase in the maximum temperatures.
3.2. O-centred ux distribution
To examine the eect of the focus point of the mirrors
drifting away from the centre of the receiver, surface tem-
peratures have been modelled using a mean focus point
50% of the distance between the centre and the side edge
of the receiver. Surfaces temperatures for the single diame-
ter receiver can be seen in Fig. 16, which shows a maximum
temperature of 943 C if control for ensuring a minimum
ow rate is not used.
To investigate the eect of o centre focusing on the
multi-diameter design, we have applied the same distribu-
tion as for Fig. 16. For this particular receiver design, it
can be seen that very high surface temperatures will result
if the aiming point for all of the heliostats moves away
from the centre of the receiver and an outlet temperature
of 570 C is still maintained (Fig. 17). The resultant maxi-
mum temperature under these conditions (1361 C) is
418 C above a single diameter design and would likely
cause damage to the receiver. As detailed, control methods
are possible to ensure that these situations do not occur,
however, the increased risk of catastrophic failure must
be considered in deciding the best receiver for a particular
application.
The eect of focus drift on surface temperatures for the
multi-pass receiver can be seen in Fig. 18. The maximum
Fig. 13. Mass ow rates required for equal exit temperatures.
Fig. 14. Surface temperatures using eleven pipe diameters.
Fig. 15. Surface temperatures for a dual-pass receiver.
N. Boerema et al. / Solar Energy 97 (2013) 356368 365
temperature is 789 C which is about 570 C lower than
using the multi-diameter design and 154 C below the sin-
gle-diameter design. This demonstrates the lower risk of
this design to incidents of high surface temperatures under
normally distributed ux intensities.
3.3. Alternative ux distribution
Whilst the predominant ux distribution on a billboard
receiver will, in general, be similar to a bell curve distribu-
tion (Vant-Hull, 1984; Schiel and Geyer, 1988; Ballestr n
and Monterreal, 2004), the occurrence of alterations to this
distribution may occur. One distribution may be where the
irradiance steps up considerably for the main portion of the
Fig. 16. Surface temperatures for a single diameter receiver with aiming
point 50% of the distance between the centre and the side edge of the
receiver.
Fig. 17. Surface temperatures for a multi diameter receiver with aiming
point 50% of the distance between the centre and the side edge of the
receiver.
Fig. 18. Surface temperatures with aiming point 50% of the distance
between the centre and the side edge of the receiver.
Fig. 19. Table top ux distribution across the aperture of the multi-pass
receiver. Stepping in irradiance from 0.5 MW/m
2
to 1 MW/m
2
for the
central 2/3 of the receiver. Vertical lines represent the edge between rst
and second pass.
Fig. 20. Surface temperatures for a multi-pass receiver using a table top
ux distribution. The outside headers each contain 15 tubes (1/4 of the
receiver).
366 N. Boerema et al. / Solar Energy 97 (2013) 356368
receiver-leading to a more table-top shaped distribution.
This may occur if the heliostats in close proximity to the
receiver, which have a lower variance, were focused using
a uniform aiming point strategy across a portion of the
receiver. This has been modelled for the multi-pass receiver
using an idealised table-top ux distribution that steps
from 0.5 MW/m
2
to 1 MW/m
2
for the central two-thirds
of the receiver. Whilst this is an idealised case, the aim is
to achieve an understanding of the inuences of a table-
top ux distribution (see Figs. 19 and 20).
The result is that as the step in irradiance does not occur
between the tubes on the edges of the headers, higher tem-
peratures occur for some of the tubes. However, when
looking at the average surface temperatures of the tubes
for the outer headers the maximum dierence is 84.4 C
which equates to a dierence in length from thermal expan-
sion of just over 2 mm.
4. Conclusion
Resultant surface temperatures for four dierent tubular
billboard designs have been presented. The eect of each
design under o-centre focus of the heliostat eld and
under a table-top ux intensity distribution has also been
examined. The high variation in HTF outlet temperatures
for the single-diameter receiver resulted in high surface
temperatures, with a maximum of 861 C. The ideal ow
receiver can reduce surface temperatures however a cost
eective method to achieve this must be demonstrated.
The multi-diameter receiver partially reduced surface tem-
peratures; however, it increased the risk of possible high
temperature events. Surface temperatures above 1300 C
resulted when the focus point was shifted away from the
receivers centre while trying to maintain a constant receiver
header outlet temperature. The multi-panel receiver, while
having greater design complexity, was shown to both
reduce surface temperatures and reduce the impact of
non-standard irradiation conditions. Under a table-top dis-
tribution, this receiver design also maintained compara-
tively low maximum surface temperatures. An increase in
the temperature dierences between tubes connected to
the same header was apparent; however, these dierences
are still considerably lower compared to those on the single
diameter receiver under a standard ux distribution. As the
maximum stress, corrosion rates and crack propagation
will all be reduced with a reduction in surface temperatures,
and as the number of cycles to failure is increased with a
reduction in the maximum stress this modelling demon-
strates the potential improvement that can be achieved
through using optimised receiver designs, where the cost/
benet must be analysed. Furthermore, the surface temper-
ature distributions and maximum temperatures will assist
in receiver material selection to achieve the design lifetime.
The maximum surface temperatures for both on and o
design positioning of the ux aim point, along with receiver
eciencies and the dierences in thermal expansion calcu-
lated using pipe average temperatures are summarised for
the discussed receiver concepts in Table 3 below.
References
A

vila-Mar n, A.L., 2011. Volumetric receivers in solar thermal power


plants with central receiver system technology: a review. Solar Energy
85 (5), 891910.
Ballestr n, J., Monterreal, R., 2004. Hybrid heat ux measurement system
for solar central receiver evaluation. Energy 29 (5), 915924.
Bignon, M.J., 1980. The inuence of the heat transfer uid on the receiver
design. Electric Power Systems Research 3 (12), 99109.
Bobinecz, M., 2012. Ivanpah solar electric generating facility. In:
Engineering and Construction Contracting Conference, San Antonio,
Texas, US, BrightSource.
Boerema, N., Morrison, G., et al., 2012. Liquid sodium versus Hitec as a
heat transfer uid in solar thermal central receiver systems. Solar
Energy (0).
Cengel, Y.A., 2007. Heat and Mass Transfer: A Practical Approach, third
ed. McGraw-Hill, America.
Collado, F.J., 2009. Preliminary design of surrounding heliostat elds.
Renewable Energy 34 (5), 13591363.
Davis, J., 2001. ASM Specialty Handbook

. Stainless Steels. ASM


International, Materials Park, OH, 1996, p. 290.
Dunn, R.I., Hearps, P.J., et al., 2012. Molten-salt power towers: newly
commercial concentrating solar storage. Proceedings of the IEEE 100
(2), 504515.
Eduardo, Z., Manuel, R.-A., 2007. Concentrating solar thermal power. In:
Handbook of Energy Eciency and Renewable Energy. CRC Press,
21-21-21-98.
Fink, J.K., Leibowitz, L., 1995. Thermodynamic and Transport Properties
of Sodium Liquid and Vapour. Reactor Engineering Division,
Argonne National Laboratory.
Haaland, S.E., 1983. Simple and explicit formulas for the friction factor in
turbulent pipe ow. Journal of Fluids Engineering, 8990.
Hinkley, J., Curtin, B., et al., 2011. Concentrating Solar Power-Drivers
and Opportunities for Cost-Competitive Electricity. Commonwealth
Scientic and Industrial Research Organisation (CSIRO), Australia.
IEA, 2010. Technology Roadmap. International Energy Agency. Con-
centrating Solar Power.
Table 3
Summary of results for the dierent billboard receiver concepts.
Receiver concept Eciency (%) Centre aim point 0.75Width aim point Thermal expansion dierence (mm)
Maximum surface temperature (C)
Single diameter 91.2 861 943 8.3
Ideal ow 91.6 634 n/a 1.5
Even ux 91.8 632 n/a 0
Multi-diameter 92.6 707 1361 5.9
Multi-pass 92.0 695 789 2.6
N. Boerema et al. / Solar Energy 97 (2013) 356368 367
IT Power, 2012. Realising the Potential of Concentrating Solar Power in
Australia. Australian Solar Institute, <http://www.australiansolarin-
stitute.com.au/reports/.aspx>.
Kesselring, P., Selvage, C.S., 1986. The IEA/SSPS Solar Thermal Power
Plants:-Facts and Figures-: Final Report of the International Test and
Evaluation Team (ITET). Vol. 1; Central Receiver System (CRS). Vol.
4; Book of Summaries. Springer.
Kolb, G.J., 2011. An Evaluation of Possible Next-generation
High Temperature Molten-salt Power Towers. Sandia National
Laboratories.
NREL, 2011. Concentrating Solar Power Projects. Retrieved 23rd
October, 2012, <http://www.nrel.gov/csp/solarpaces/power_tower.
cfm>.
Romero, M., Buck, R., et al., 2002. An update on solar central receiver
systems, projects, and technologies. Journal of Solar Energy Engi-
neering 124 (2), 98108.
Schell, S., 2011. Design and evaluation of esolars heliostat elds. Solar
Energy 85 (4), 614619.
Schiel, W.J.C., Geyer, M.A., 1988. Testing an external sodium receiver up
to heat uxes of 2.5 MW/m
2
: results and conclusions from the IEA
SSPS high ux experiment conducted at the central receiver system of
the Plataforma Solar de Almeria (Spain). Solar Energy 41 (3), 255265.
Vant-Hull, L., 1984. Instantaneous ux distribution on a solar central
receiver. Journal of Solar Energy Engineering; (United States) 106 (1).
Viswanathan, R., 1989. Damage Mechanisms and Life Assessment of
High-temperature Components. ASM International, OH.
Ward, I., 2012. New industrial trends in high temperature alloys for
concentrating solar thermal systems. In: Concentrating Solar Thermal
Material Science and Engineering. Sydney, Australia, Sandvik Mate-
rials Technology.
Yang, X., Yang, X., et al., 2012. Numerical simulation study on the heat
transfer characteristics of the tube receiver of the solar thermal power
tower. Applied Energy 90 (1), 142147.
Zhang, J.J., Pye, J.D., et al., 2012. Estimation of uncertainty in automated
heliostat alignment. In: Australian Solar Council Solar 2012 Confer-
ence. Melbourne.
368 N. Boerema et al. / Solar Energy 97 (2013) 356368

Das könnte Ihnen auch gefallen